INDEX

Explanations

calculation of parts of words

The neuron fires on technical/math‐related operation words (e.g. “calculate,” “function(s),” “derivative”) indicating computational instructions or concepts.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 That

-1.56

 THAT

-1.50

那个

-1.48

The

-1.45

THE

-1.41

 той

-1.37

 splits

-1.29

That

-1.22

-1.21

mış

-1.20

POSITIVE LOGITS

they

1.50

只需要

1.48

ㄟ

1.34

fusions

1.34

くらいで

1.34

âmetros

1.33



1.32

⦊

1.31

 ので

1.28

niños

1.26

Activations Density 0.044%